A method and system for controlling the moisture content of a fabric on a stenter based on MPC

Abstract

The present application relates to the technical field of textile printing and dyeing, and particularly relates to a setting machine cloth drop moisture content control method and system based on MPC, which comprises collecting control parameters; constructing an MPC model, the MPC model comprising: taking the measured and preset moisture content and a first control parameter state vector as inputs to obtain a predicted moisture content; taking the minimum difference between the predicted moisture content and the preset moisture content as a target; using an optimization matrix and a sequential quadratic programming to solve and optimize the control increment; and using the optimal control increment to control the action of an actuator. The present application solves the problems of insufficient detection accuracy, hysteresis and poor anti-interference of the existing single temperature feedback control fabric moisture content.

Description

Technical Field

[0001] This invention relates to the field of textile printing and dyeing technology, and in particular to a method and system for controlling the moisture content of fabric dropped from a setting machine based on MPC. Background Technology

[0002] In the process of fabric finishing and dyeing, it is crucial to accurately control the moisture content of the finished fabric in order to ensure the hand feel, dimensional stability, color uniformity and other physical properties of the fabric in subsequent processing or use.

[0003] Existing drying and setting equipment controls fabric moisture content through single temperature feedback, which has the following drawbacks:

[0004] 1. Lag: Relying solely on the drying room temperature to indirectly estimate the fabric moisture content cannot respond in real time to actual changes in fabric moisture content;

[0005] 2. Insufficient precision: The nonlinear relationship between temperature and moisture content is complex, and controlling with a single temperature threshold can easily lead to over-drying or under-drying;

[0006] 3. Poor resistance to interference: Factors such as fluctuations in environmental humidity, differences in fabric type and structure are not dynamically compensated. Summary of the Invention

[0007] To address the shortcomings of existing methods, this invention utilizes an MPC model for multi-sensor data fusion and coordinates it with actuators for control, thereby achieving precise adjustment of fabric moisture content.

[0008] The technical solution adopted in this invention is: a method for controlling the moisture content of fabric dropped from a stenter based on MPC, comprising the following steps:

[0009] Step 1: Collect control parameters;

[0010] In a preferred embodiment of the present invention, the control parameters include: temperature value, humidity value, water content, air pressure difference value, vehicle speed value, exhaust volume, and the ratio of heat source power to circulating fan flow rate.

[0011] Step 2: Construct the MPC model. The MPC model includes: taking the measured and preset moisture content and the first control parameter state vector as inputs to obtain the predicted moisture content; minimizing the difference between the predicted moisture content and the preset moisture content as the objective; solving and optimizing the control increment using the optimization matrix and sequential quadratic programming; and using the optimal control increment to regulate the action of the actuator.

[0012] In a preferred embodiment of the present invention, the first control parameters include: average humidity of the drying room, average temperature of the drying room, vehicle speed, and the ratio of heat source power to circulating fan flow rate.

[0013] In a preferred embodiment of the present invention, step two specifically includes:

[0014] Preset prediction time domain N p Control time domain N c W H and W u ;

[0015] Construct the prediction equation:

[0016] Among them, H pred (k+1) is the predicted moisture content at time k+1; H(k) is the measured moisture content of the fabric at the point where the fabric is laid at time k; α is the temperature influence coefficient; β is the humidity self-regulation coefficient; γ is the hot air ratio coefficient; ΔT is the difference in average temperature of the drying room at adjacent times; ε is the adjustable coefficient; v is the vehicle speed; Q / F is the ratio of heat source power to circulating fan flow rate;

[0017] Construct the minimum objective function:

[0018] Among them, W H To weigh the error weights; W u The smoothness of the control increment is weighed; Δu is the control increment.

[0019] Construct an optimization matrix and use the SQP algorithm to solve for Δu.

[0020] As a preferred embodiment of the present invention, the optimized matrix formula is:

[0021]

[0022] Where A is the first Jacobian matrix, b H b u Let ΔX be the vector of error term and control increment term, and let ΔX represent the increment of state vector X(k).

[0023] In a preferred embodiment of the present invention, the optimization of the control increment adopts an online correction model of model parameters, and the formula is: θ(k'+1)=θ(k')+μ·J －1 (θ(k'))*(H(k')-H pred (k')), where μ is the adaptive learning rate, θ k’ Let J be the control parameter coefficient vector at the k'-th iteration, J-1 be the inverse of the second Jacobian matrix, and H(k') be the measured moisture content of the fabric at the placement point at the k'-th iteration. pred (k') Predicted water content at the k'+1th iteration.

[0024] In a preferred embodiment of the present invention, the control increment includes: the change in the rotational speed of the vehicle speed motor, the change in the flow rate of the circulating fan at adjacent time intervals, the change in the power of the heat source at adjacent time intervals, and the change in the exhaust volume at adjacent time intervals.

[0025] In a preferred embodiment of the present invention, the control increment is constrained by the humidity gradient between adjacent drying rooms.

[0026] In a preferred embodiment of the present invention, the control increment is constrained by the air pressure difference between adjacent drying rooms.

[0027] As a preferred embodiment of the present invention, a system employing an MPC-based method for controlling the moisture content of fabric dropped from a setting machine includes:

[0028] The actuator, controller, humidity sensor, temperature sensor, and moisture content sensor are included. The actuator is electrically connected to the controller, and the humidity sensor, temperature sensor, and moisture content sensor are also electrically connected to the controller.

[0029] in,

[0030] Humidity sensors are used to collect humidity values ​​in each drying room;

[0031] Temperature sensors are used to collect temperature values ​​inside each drying chamber;

[0032] Moisture content sensors are used to detect the moisture content of fabric at the point where it is laid.

[0033] The controller uses the collected humidity, temperature, and moisture content values, along with a trained MPC model, to control the actuator's actions to adjust the moisture content.

[0034] The beneficial effects of this invention are:

[0035] 1. Construct an MPC model and use it to collect indicators such as temperature, humidity, moisture content, and air pressure inside and outside the drying room. Use the MPC model to fine-tune the actions of the actuator to achieve precise control of the fabric moisture content; achieve a moisture content control accuracy of ±0.5% to 2%, which is significantly better than the ±3% to 5% error of traditional single temperature feedback.

[0036] 2. Adaptive adjustment of prediction time-domain Np and control time-domain Nc, combined with weighting matrix W H W u Dynamic optimization compensates for disturbances such as fluctuations in environmental humidity, fabric type, and differences in structure in real time to ensure process stability;

[0037] 3. Construct an online correction model for model parameters, iteratively optimize the control parameter coefficients, and obtain the optimal control increment;

[0038] 4. Optimize humidity gradient constraints by adapting different processes to cotton and synthetic fibers, setting gradient constraints to avoid uneven fiber expansion due to sudden humidity changes, thus reducing the risk of wrinkles and deformation; reduce energy waste in adjacent drying chambers through humidity gradient constraints, combined with dynamic matching of machine speed and heat source power; process 200g / m³... 2 When processing cotton fabrics, humidity gradient constraints reduce the average temperature of the drying chamber from 185℃ to 178℃, resulting in a 12% reduction in energy consumption. For a 5-section drying chamber setting machine processing synthetic fiber fabrics, a gradient constraint optimizes the humidity distribution from 60%→45%→30%→15%→5%RH to 55%→45%→35%→25%→15%RH. Combined with dynamic matching of machine speed (10-100m / min) and heat source power (20-100%), this achieves optimal processing for 200g / m³ fabrics. 2 When processing cotton fabrics, gradient constraints can reduce the average temperature of the drying room from 185℃ to 178℃, which can reduce the frequency of fan and heat source power adjustment and reduce energy consumption by 12%.

[0039] 5. Pressure difference ΔP between adjacent drying rooms i,i+1 ≤50Pa ensures uniform airflow distribution and avoids excessively fast or slow airflow speeds, which can affect the uniformity of moisture evaporation on the fabric, resulting in insufficient or excessive drying of certain parts of the fabric and affecting the quality of the final product. For example, at a speed of 100m / min, differential pressure control can reduce the transverse moisture content deviation of the fabric from ±2% to ±0.8%. Attached Figure Description

[0040] Figure 1 This is a flowchart of the method for controlling the moisture content of fabric dropped into the stenter based on MPC according to the present invention;

[0041] Figure 2 This is a schematic diagram of the moisture content control system for the fabric falling off the MPC stenter of the present invention. Detailed Implementation

[0042] The present invention will be further described below with reference to the accompanying drawings and embodiments. The drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0043] like Figure 1 As shown, a method for controlling the moisture content of fabric dropped from a stenter based on MPC includes the following steps:

[0044] Step 1: Collect control parameters;

[0045] The control parameters include: temperature, humidity, moisture content, air pressure difference, vehicle speed, exhaust volume, and the ratio of heat source power to circulating fan flow rate.

[0046] It also includes: eliminating outliers in control parameters, which can be done using the 3σ criterion.

[0047] The temperature values ​​are collected using several temperature sensors arranged along the direction of fabric movement in the drying room.

[0048] Temperature sensor T n n is the number of temperature sensors, n>1;

[0049] The temperature sensor can be a distributed fiber optic temperature sensor, model IF-DTS. The temperature sensor is placed inside the drying chamber and is fixedly connected to the drying chamber and electrically connected to the controller.

[0050] The temperature sensor can also be an armored K-type thermocouple, which measures the temperature of the space and is inserted into the inner cavity of the drying chamber. The thermocouple is fixedly connected to the drying chamber and electrically connected to the controller.

[0051] The humidity value is collected using several humidity sensors arranged along the direction of fabric movement in the drying room.

[0052] Humidity sensor Hair m m is the number of humidity sensors, m>1;

[0053] The humidity sensor can be a high-temperature capacitive humidity sensor; it is placed on the inner wall of the drying chamber, with an accuracy of ±3% to 5% RH, and is fixedly connected to the drying chamber and electrically connected to the controller.

[0054] The humidity sensor can also be an optical humidity sensor, placed inside the drying chamber with the optical window facing the main airflow direction, with an accuracy of ±1.5% to 3% RH. It is fixedly connected to the drying chamber through a shock absorber and electrically connected to the control layer.

[0055] Among them, the moisture content is collected by a fabric moisture content sensor at the fabric drop point of the setting machine;

[0056] The moisture content sensor MC can be an infrared absorption sensor or a capacitive sensor, placed on the frame of the fabric dropping point of the setting machine outside the drying oven. The measurement accuracy is ±0.5%~2% / ±2%~5%. It is fixedly connected to the drying oven and electrically connected to the controller.

[0057] The moisture content sensor can also be a microwave moisture content sensor, which is placed on the frame of the setting machine outside the drying room to measure the moisture content of the fabric with an accuracy of ±0.5%. It is fixedly connected to the drying room and electrically connected to the controller.

[0058] The air pressure difference value is collected by using air pressure sensors installed between each drying chamber of the stenter machine.

[0059] The air pressure difference ΔP is the air pressure difference between adjacent drying rooms. i,i+1 =PA i-PA i+1 , accuracy ±0.5% FS;

[0060] Pressure sensor PA w w is the number of humidity sensors, w>1;

[0061] The pressure sensor can be a high-temperature resistant ceramic capacitive / thermocouple differential pressure sensor, placed inside the drying chamber, with an accuracy of ±0.5%FS / ±1.0%FS, fixedly connected to the drying chamber and electrically connected to the controller;

[0062] The pressure sensor can also be a high-temperature piezoresistive sensor, placed outside the drying chamber, connected to the inner cavity of the drying chamber through a pressure-sensing tube that has been insulated / heat-traced, and is fixedly connected to the drying chamber and electrically connected to the controller.

[0063] Vehicle speed, circulating fan flow rate, heat source power, and exhaust volume are obtained using actuators, which include: vehicle speed motor, variable frequency circulating fan, heat source proportional valve, and electric exhaust valve;

[0064] The speed motor is mainly used to pull the fabric on the fabric clips / needle plate of the setting machine. Its speed value v is usually between 10 and 100 m / min to meet the setting requirements of different fabrics. The speed value v is directly proportional to the speed M of the speed motor.

[0065] Variable frequency circulating fans ensure uniform flow of hot air inside the setting machine. The operating frequency range of the variable frequency circulating fan is between 30 and 60 Hz. By adjusting the fan speed through the frequency converter, precise control of the hot air flow rate inside the drying chamber can be achieved, thereby ensuring that the temperature and humidity inside the setting machine are at the optimal level. The output circulating fan flow rate F can be adjusted by changing the circulating fan speed v'.

[0066] The heat source proportional valve is used to control the heating system of the stenter. The heat source proportional valve of the stenter is installed on the heat source delivery pipeline, such as a steam pipeline or a gas pipeline. Its adjustment range is between 20% and 100%. By adjusting the opening of the proportional valve, the heat output of the heating system can be precisely controlled; the output heat source power value Q is also controlled.

[0067] The electric exhaust valve is driven by a motor to open and close, with an opening range of 0 to 100%. It is used to control the airflow between the drying room and the external environment, and to adjust the airflow inside the setting machine, thereby ensuring that the temperature, humidity and pressure inside the drying room are at the optimal state, and outputs an exhaust volume E.

[0068] It also includes: setting fabric property parameters;

[0069] Fabric property parameters include: fabric weight (g / m³) 2 (Accuracy ±1g / m) 2Fabric width (m), accuracy ±0.01m; fabric moisture content preset value (%), accuracy ±0.5%; fabric type includes: cotton or chemical fiber.

[0070] Step 2: Construct the MPC model. The MPC model includes: taking the measured and preset moisture content and the first control parameter state vector as inputs to obtain the predicted moisture content; minimizing the difference between the predicted moisture content and the preset moisture content as the objective; solving and optimizing the control increment using the optimization matrix and sequential quadratic programming; and using the optimal control increment to regulate the action of the actuator.

[0071] The first control parameter includes: average humidity of the drying room. avg Average temperature of the drying room (T) avg Vehicle speed v, heat source power to circulating fan flow rate Q / F;

[0072] Among them, Hair avg T represents the average value from m humidity sensors. avg The average value is the value from n temperature sensors.

[0073] The formula for the first control parameter state vector is:

[0074] X(k) = [Hair] avg (k),T avg (k),v(k),Q / F(k)] T (1)

[0075] Where k is the time; X(k) is updated periodically, and in this embodiment, the time interval Δt = 5 seconds;

[0076] The first control parameter state vector is 4-dimensional, which is a mathematical abstraction of the thermodynamic state and equipment operating parameters inside the setting machine drying chamber. Each dimension represents a key variable affecting the moisture content of the fabric exiting the setting machine:

[0077] Hair avg (k) represents the average air humidity (%RH) inside the drying room at time k, reflecting the water vapor content inside the drying room;

[0078] T avg (k) represents the average temperature (°C) inside the drying room at time k, which determines the thermodynamic driving force for water evaporation.

[0079] v(k) is the vehicle speed (m / min) at time k, which is fed back in real time by the vehicle speed motor and affects the dwell time of the fabric in the drying room;

[0080] Q / F(k) is the ratio (kW / Hz) of the heat source power Q to the circulating fan flow rate F at time k, reflecting the thermal intensity of the hot air;

[0081] These four variables constitute the core state parameters of the drying process in the drying room. By combining them linearly or nonlinearly, a state-space model is established to characterize the dynamic changes of the system (such as drying room humidity, drying room temperature, vehicle speed, heat source power and circulating fan flow ratio) over time, providing basic input for the MPC prediction model.

[0082] The construction of an MPC model includes:

[0083] 1. Preset prediction time domain N p Control time domain N c Weight matrix W H and W u ;

[0084] In this embodiment, N p =10; N c =5; W H =diag(0.6,0.2,0.2), where W represents the weights for moisture content error, temperature error, and humidity error, respectively. u =diag(0.1,0.15,0.1,0.05), which are the smoothness weights of vehicle speed, circulating fan speed, heat source power, and drying room exhaust opening, respectively;

[0085] It also includes: setting control parameter constraints; among which,

[0086] Humidity gradient constraints include:

[0087] Humidity gradient constraint for cotton fabrics: Humidity gradient constraint threshold between adjacent drying rooms | Hair i -Hair i+1 |≤15%;

[0088] Humidity gradient constraint for synthetic fiber fabrics: Humidity gradient constraint threshold between adjacent drying rooms | Hair i -Hair i+1 |≤10%;

[0089] When constructing the optimization problem, the MPC model incorporates the humidity gradient constraint as a hard constraint in the state space model into the objective function solution process. The humidity gradient constraint directly affects the construction of the average humidity of the drying room in the state space model by restricting the humidity variable. When the humidity difference between adjacent drying rooms approaches or exceeds the constraint threshold at a certain moment, the model will adjust the prediction of the future state, because abnormal changes in humidity may cause deviations in the control of fabric moisture content, thereby affecting the operating state of the entire system.

[0090] The objective function of Equation 3 in the MPC model aims to make the actual moisture content of the fabric at the setting machine drop point as close as possible to the moisture content H set by the process, while satisfying a series of constraints. setThis reduces frequent changes in the actuator output; the humidity gradient constraint is embedded as a hard constraint in the solution process of this objective function; during optimization calculation, the algorithm predicts the fabric moisture content at multiple future times based on the current system state and control input at each sampling time; in this prediction process, the humidity gradient constraint limits the range of humidity variable changes to ensure that the predicted humidity state meets the constraint requirements; if the humidity gradient in the prediction result exceeds the constraint range, the controller adjusts the actuator output to meet the humidity gradient constraint and make the objective function optimal.

[0091] For example, the MPC model executes cyclically, continuously detecting whether the humidity gradient constraint meets the threshold condition. When the threshold is exceeded, W is adjusted. H The humidity error weights are adjusted to increase ΔE and decrease ΔM in the MPC model.

[0092] For example, if the humidity of the chemical fiber fabric suddenly increases in the third drying room, the humidity gradient constraint value will be 12%, which is 2% over the limit. The MPC model will increase the opening of the exhaust valve ΔE (maximize within the limit) or increase ΔF, while slightly reducing ΔM (extend the residence time) to quickly restore the humidity gradient balance.

[0093] In addition, the humidity gradient constraint also affects the average humidity of the drying room in X(k). avg When the humidity difference between adjacent drying rooms exceeds a threshold, the system determines that the current humidity distribution is abnormal and triggers Equation 5, which compares the measured moisture content H(k) with the predicted value H pred To correct the deviation of (k), the humidity self-regulation coefficient β in Formula 2 is adjusted (e.g., β is increased from 0.12 to 0.14), thereby increasing the weight of the influence of humidity on moisture content prediction.

[0094] For example, in cotton fabric production, if the humidity of the second drying room is 60% and the humidity of the third drying room is 40% (humidity gradient 20% > 15%), the MPC model will force the adjustment of the opening degree E of the exhaust valve of the third drying room, and at the same time increase the humidity self-adjustment coefficient β in the prediction equation of Equation 2 to suppress the risk of subsequent gradient exceeding the standard.

[0095] In dynamic optimization, when the humidity gradient constraint conflicts with other constraints (such as temperature safety boundaries and vehicle speed limits), the MPC model follows the process priority principle:

[0096] Low-priority scenarios: When the temperature does not exceed the limit, prioritize satisfying the humidity gradient constraint (e.g., sacrifice some heat source power adjustment speed to ensure gradient ≤ 15%).

[0097] High-risk scenario: When the humidity gradient continuously exceeds the standard and is accompanied by abnormal temperature, the "humidity gradient + temperature" dual constraint is triggered, and the exhaust valve E and heat source Q are adjusted simultaneously to avoid fiber damage (e.g., when the humidity gradient of chemical fiber fabric exceeds the standard + temperature > 150℃, the vehicle speed is automatically reduced and the exhaust is increased, providing dual protection).

[0098] For fabrics made from different fiber raw materials, the humidity gradient constraint between adjacent drying chambers must be adjusted accordingly. For example, cotton fabrics are more sensitive to humidity changes and require a smaller humidity gradient constraint, while synthetic fiber fabrics are less sensitive to humidity changes and can tolerate a larger humidity gradient constraint. This helps to meet the processing needs of different types of fabrics. The humidity gradient constraint in the drying chamber is an important innovation of this invention, and it has significant implications for dyeing and finishing processes. Specifically:

[0099] 1. Improve fabric quality: By controlling the humidity gradient between adjacent drying rooms, problems such as wrinkles and deformation caused by excessive humidity changes during the setting process can be avoided. This helps to ensure the physical properties of the fabric, such as hand feel and dimensional stability, in subsequent processing or use.

[0100] 2. Optimize energy consumption: Reasonable humidity gradient constraints can reduce the humidity difference between drying rooms, thereby reducing energy consumption. For example, when the humidity gradient is large, it is necessary to increase the heat source power or extend the drying time to ensure that the fabric reaches the ideal moisture content. By controlling the humidity gradient, energy consumption can be reduced while ensuring the quality of the fabric.

[0101] It also includes: the air pressure difference constraint threshold ΔP between adjacent drying rooms. i,i+1 =PA i -PA i+1 ≤50Pa;

[0102] Changes in the air pressure difference ΔP alter the airflow pattern within the drying chamber, thus affecting the average humidity H of the drying chamber. avg And Q / F; when ΔP increases, it may cause abnormal airflow velocity in the drying oven, altering the moisture distribution and affecting H. avg Changes in airflow can also affect heat transfer efficiency, which in turn affects the coordinated operation of the heat source power and the circulating fan flow, leading to changes in Q / F. These changes ultimately affect the actual state of the drying room reflected by the state vector. ΔP indirectly affects the calculation of the Q / F term in Equation 2, thereby changing the effect on the fabric moisture content H at the next moment. pred The predicted value enables the prediction model to more accurately reflect the actual situation; in the optimization solution of Equation 4, ΔP≤50Pa is used as a constraint condition; when the predicted H pred With H setWhen an error exists and ΔP approaches or exceeds the limit, it will constrain the adjustment of the control increment. If ΔP is too large, when solving Δu, the algorithm will prioritize adjusting control variables related to air pressure, such as the opening degree of the exhaust valve ΔE and the speed of the circulating fan v', in order to meet the air pressure difference constraint and balance the moisture content error, so as to ensure that the objective function reaches the optimal solution and achieve precise control of the fabric moisture content and stable system operation.

[0103] Also includes: maximum permissible temperature T max The constraints are: 180℃ for cotton and 150℃ for synthetic fibers.

[0104] 2. Construct the prediction equation using the control parameters. The formula is as follows:

[0105]

[0106] Among them, H pred (k+1) is the predicted moisture content at time k+1; H(k) is the measured moisture content of the fabric at the point where it is laid at time k; α is the temperature influence coefficient; β is the humidity self-regulation coefficient; γ is the hot air ratio coefficient; ΔT is the difference in average temperature of the drying room at adjacent times, ΔT = T avg (k+1)-T avg (k); ε is the adjustable coefficient; v is the vehicle speed; Q / F is the ratio of heat source power to circulating fan flow rate;

[0107] In this embodiment, α = 0.025–0.035, β = 0.10–0.15, γ = 0.005–0.015, and v = 0.5; α is dynamically adjusted according to the fabric weight; β is adaptively adjusted according to the fluctuation of ambient humidity; and γ is determined by regression from historical data.

[0108] You can also set constraints, including: speed change limits and exhaust valve step limits.

[0109] Vehicle speed change constraint: |Δv|≤3m / min, dead zone ±0.5m / min;

[0110] Exhaust valve step constraint: |ΔE|≤8% / Δt (Δt=5s);

[0111] Temperature safety boundary constraints: T avg <T max If it exceeds T max Trigger an alarm and reduce the heat source power Q;

[0112] ΔT reflects the dynamic changes in the drying room temperature and has a significant impact on the evaporation of moisture from the fabric; v affects the residence time of the fabric in the drying room, thus affecting the drying effect.

[0113] Q / F reflects the heat intensity of the hot air and affects the rate at which moisture evaporates from the fabric.

[0114] α quantifies the degree of influence of temperature change ΔT on the change of fabric moisture content. The value of α varies depending on the type of fabric and the drying environment.

[0115] β is used to measure the relationship between H(k) and H. set The impact of the differences between them on future moisture content changes;

[0116] γ reflects the magnitude of the effect of Q / F on the moisture content of the fabric;

[0117] α·ΔT·v ﹣ε The larger the temperature change ΔT, the faster the fabric moisture evaporates and the greater the decrease in moisture content. ΔT is positively correlated with the change in moisture content.

[0118] v ﹣ε This indicates that the faster the machine speed, the shorter the time the fabric stays in the drying room, and the less chance there is for moisture to evaporate per unit time, so the decrease in moisture content is relatively small; this factor comprehensively considers the effects of temperature and machine speed on moisture content changes.

[0119] Hit when H(k)>H set hour, At this point, the term is negative, meaning the moisture content will decrease towards the target value; when H(k) <H set hour, If this value is positive, it indicates that the moisture content may increase to approach the target value. This value reflects that the system of the present invention has a mechanism to automatically adjust the moisture content to approach the target value.

[0120] In γ·Q / F(k), the larger Q / F is, the higher the heat intensity of the hot air, the faster the moisture in the fabric evaporates, and the more obvious the decrease in moisture content. Therefore, this term is positively correlated with the change in moisture content.

[0121] 3. Optimize control input. The core of MPC is solving an optimization problem, namely minimizing the objective function. The formula for minimizing the objective function is:

[0122]

[0123] Among them, W H To weigh the error weights, such as moisture content error, temperature error, and humidity error; W u Weigh the smoothness of the control increment; Δu is the control increment, Δu = [ΔM, ΔF, ΔQ, ΔE] T ΔM is the change in the speed of the vehicle speed motor, ΔF is the change in the flow rate of the circulating fan at adjacent time points, ΔQ is the change in the power of the heat source at adjacent time points, and ΔE is the change in the exhaust air at adjacent time points.

[0124] Formulas 2 and 3 are the core of the moisture content prediction model for the fabric falling off the stenter in this invention. The system can predict H based on the current H(k), ΔT, v, and Q / F information. pred (k+1); H pred (k+1) is used for the optimization solution of formula (3), H pred With H set The error is calculated to determine the optimal control input change Δu = [ΔM, ΔF, ΔQ, ΔE]. T Therefore, explicitly predicting N in the time domain p and control time domain N c The value of is taken, and the future N is calculated based on the state-space model and the current state. p The predicted value H at each time point pred (k); Prediction time domain N p This determines the range of the summation of the error terms, i.e. The upper limit of summation in the time domain; controlling the time domain N c This is related to the summation range of the control input changes, i.e. The upper limit of the summation can be set; at the same time, some constraints can be set to limit the value of the control variable Δu in the optimization solution.

[0125] 4. Construct the optimization matrix and use the Sequential Quadratic Programming (SQP) algorithm to solve for Δu;

[0126] The formula for optimizing the matrix is:

[0127]

[0128] Where A is the first Jacobian matrix, b H b u Let ΔX be the vector of error term and control increment term, and let ΔX represent the increment of state vector X(k), which is the difference between the current state and the next state in MPC, i.e.

[0129] Real-time computing Iso-partial derivatives; partial derivatives It represents the rate of change of fabric moisture content with vehicle speed, that is, the degree of influence of the change of vehicle speed v on fabric moisture content. It is a key element of A and is used to quantify the dynamic influence of the control variable (vehicle speed) on the output variable (fabric moisture content).

[0130] Similarly, real-time computing Together they form the Jacobian matrix, which is used to quantify the relationship between control variables and output variables;

[0131] The velocity term "α·ΔT·v" in Equation 2 -0.5 “” indicates the contribution of vehicle speed v to the change in moisture content;

[0132] Partial derivatives It is a key element of A, which is used to describe control variables, such as vehicle speed v, fan flow rate F, heat source power Q, and exhaust opening degree E, which affect the fabric moisture content H. pred The dynamic effects; for example: when the vehicle speed v increases, The absolute value decreases (because v) ﹣1.5 The decreasing function indicates that the influence of vehicle speed on moisture content weakens at high speeds, and the algorithm will automatically adjust the control strategy to avoid over-adjustment.

[0133] Let V be the first-order partial derivative of the velocity term in Equation 2 with respect to the vehicle speed v, i.e.: The two represent the mathematical relationship between the antiderivative and the derivative;

[0134] In Equation 2, α·ΔT·v -0.5 This indicates that the faster the machine speed (v), the shorter the time the fabric stays in the drying room, resulting in less moisture evaporation and a smaller decrease in moisture content. This is indicated by the speed term (v). -0.5 The nonlinear effect of vehicle speed on moisture content: Vehicle speed is negatively correlated with changes in moisture content, but the degree of influence weakens as vehicle speed increases;

[0135] The velocity term describes the nonlinear effect of vehicle speed on moisture content changes, and the partial derivative quantifies the instantaneous rate of this effect. Together, they constitute the core of "model prediction" and "optimization solution" in the MPC model. This deep coupling between mathematics and physics enables the system to dynamically adapt to complex working conditions and realize closed-loop control of "model prediction-optimization solution-execution adjustment". This is the key technical advantage of this invention that distinguishes it from traditional single-variable control (such as temperature control only).

[0136] SQP solution includes: setting an upper limit for the number of iterations, which is 100; or setting a convergence threshold of 1e-4 or 1e-5.

[0137] The optimal control increment is Δu = [ΔM, ΔF, ΔQ, ΔE]. T ;

[0138] The optimized control increment is transmitted to the actuator to achieve real-time control; this is a direct application of the MPC algorithm output, directly corresponding to the optimization solution part of the MPC algorithm; according to H pred (k), W H and W u Formula (3) is constructed as the objective function; and nonlinear optimization algorithms (such as SQP, IPOPT, etc.) are used to solve the objective function to obtain the optimal control input change Δu, so that the objective function reaches the minimum value.

[0139] During the solution process, W H and W uIt affects the weights of the error term and the control input change term, thereby balancing the system's requirements for error control and control input changes; the optimized control increment is then transmitted to the actuator to achieve real-time control.

[0140] It also includes: constructing online correction of model parameters, iterating the control parameter coefficients of the prediction equation, and realizing closed-loop feedback;

[0141] θ(k'+1)=θ(k')+μ·J －1 (θ(k'))*(H(k')-H pred (k')) (5)

[0142] Where μ is the adaptive learning rate, μ = 0.02 to 0.1. Increasing μ (e.g., 0.1) when the error increases sharply accelerates convergence, while decreasing μ (e.g., 0.02) when the error stabilizes avoids over-adjustment; J⁻¹ is the inverse of the second Jacobian matrix, used to quantify the influence direction of each control parameter coefficient on the moisture content. For example, when the temperature increases by 1℃, α contributes 60%, guiding parameter adjustments to focus on key factors; θ k’ Let be the control parameter coefficient vector for the k'th iteration. The control parameter coefficients include α, β, and γ.

[0143] The A matrix in Equation 4 serves to optimize the solution of control variables, describing the dynamic relationship of "control action → moisture content change". The A matrix has a dimension of 1x4, the input variables are vehicle speed, fan, heat source, and exhaust, and the output is H. pred The partial derivatives of the variables of vehicle speed, fan, heat source, and exhaust are used to solve the rolling optimization problem of the MPC model, that is, to calculate the optimal control increment Δu.

[0144] In Equation 5, J is the Jacobian matrix for model parameter correction, with a dimension of 1x3. It represents the rate of influence of model parameters α, β, and γ on the output variable (i.e., the predicted water content). J focuses on the "influence of model parameters on the prediction results", A is used for the "optimization layer" of the MPC model, and J is used for the "correction layer" of the MPC model.

[0145] This invention utilizes H(k) and H pred The error signal of (k) is combined with the dynamic control parameter coefficients μ and J to improve the prediction accuracy.

[0146] This invention uses X(k) to predict the fabric moisture content H in Formula 2. pred , H pred With H set Substituting into Equation 3, we use Equation 4 to solve the problem. During the solution process, we need to consider the speed change limit Δv, the exhaust valve step limit ΔE, and the temperature safety boundary T. max Under the given constraints, the optimal control increment Δu = [ΔM, ΔF, ΔQ, ΔE] is obtained by solving the sequential quadratic programming (SQP) algorithm.T The actuator is controlled by Δu; and the model parameters in Equation 5 are corrected online. The calculation result of Equation 5 is given to Equation 2 to update the α, β, and γ of the prediction model in Equation 2, so that the model is closer to the actual drying process and the prediction accuracy is improved.

[0147] Then the next control cycle begins, and the above process is repeated.

[0148] Online correction of model parameters plays a crucial role in the practical application of MPC models. During the fabric finishing and dyeing process, the prediction accuracy of the model may be affected by factors such as fluctuations in environmental humidity, fabric structure, and type differences. Therefore, this invention constructs an online correction mechanism for model parameters to adjust the model parameters through real-time feedback, thereby improving control accuracy and stability.

[0149] It also includes: coordinated adjustment of the implementing agencies, including: incremental limit output and interlock protection mechanisms;

[0150] The incremental limiting output includes:

[0151] Vehicle speed increment ΔM: ±2m / min, dead zone ±0.5m / min;

[0152] Circulating fan flow rate increment ΔF: ±3Hz, dead zone ±1Hz;

[0153] Heat source power increment ΔQ: ±5%, dead zone ±2%;

[0154] Exhaust opening increment ΔE: ±6%, dead zone ±3%;

[0155] Interlock protection mechanism:

[0156] When the vehicle speed v < 5m / min, the heat source power Q automatically drops to the safe level (≤20%).

[0157] If ΔP i,i+1 If the pressure is >50Pa, prioritize adjusting the exhaust valve opening E and limit the rate of change of the circulating fan flow rate F.

[0158] This step applies the optimal control input change Δu obtained from Equation 4 to the actual actuator. In the MPC model of this invention, usually only the first control input change Δu in the control time domain Nc is executed. Subsequent control inputs will be recalculated in the next rolling optimization, i.e., control time domain Nc = 5, each time the control increment Δu(k) (k = 0 to 4) for the next 5 moments is calculated, but only Δu(0) is executed. Δu is recalculated in the next sampling period (after 5 seconds) to achieve dynamic update.

[0159] Meanwhile, limiting and interlocking the control input are performed to ensure the rationality of the control input and the safety of the system. This is also consistent with the constraint on the change of control input in the minimum objective function. In accordance with this, we should avoid excessive changes in the control input; the optimal amount of change in the control input should be minimized to minimize the value of the objective function.

[0160] It also includes: temperature gradient anomaly detection;

[0161] Temperature gradient anomaly detection includes: monitoring the temperature gradient of the drying oven and fault diagnosis and response;

[0162] Monitoring the temperature gradient in the drying oven includes: calculating the temperature difference ΔT between adjacent temperature sensors. i,i+1 =T i -T i+1 If ΔT i,i+1 A temperature exceeding 10℃ for 10 seconds will trigger an air duct blockage warning.

[0163] Fault diagnosis and response include: After an alert is issued, the system automatically performs the following operations:

[0164] 1. Increase the fan speed to 80% of the maximum allowable value;

[0165] 2. Gradually increase the opening of the exhaust valve, increasing it by 10% each time;

[0166] 3. If the problem is not resolved within 30 seconds, production will be paused and a maintenance check will be prompted.

[0167] 4. Record fault logs to the cloud server.

[0168] Temperature gradient anomaly detection is a feedback correction loop, meaning it detects whether the system has encountered an anomaly. When an anomaly is detected, the control strategy is adjusted, which is equivalent to feedback correction of the MPC prediction model, making subsequent predictions and optimization calculations more accurate, thereby ensuring that the system can continuously and stably move towards the target H. set run.

[0169] Recording fault logs to the cloud server provides data support for subsequent system optimization and MPC model improvement; it is of great significance for the continuous optimization and upgrading of the MPC model.

[0170] like Figure 2 A system for controlling the moisture content of fabric dropped into a setting machine based on MPC includes: an actuator, a controller, a humidity sensor, a temperature sensor, and a moisture content sensor; the actuator is electrically connected to the controller, and the humidity sensor, temperature sensor, and moisture content sensor are electrically connected to the controller.

[0171] in,

[0172] Humidity sensors are used to collect humidity values ​​in each drying room;

[0173] Temperature sensors are used to collect temperature values ​​inside each drying chamber;

[0174] Moisture content sensors are used to detect the moisture content of fabric at the point where it is laid.

[0175] The controller uses the collected humidity, temperature, and moisture content values, along with a trained MPC model, to control the actuator's actions to adjust the moisture content.

[0176] The controller can be a microcontroller or MCU controller; it can also be a new type of electronic controller (CTR). The CTR is electrically connected to the inherent electronic control device of the machine. The CTR is an industrial control computer (IPC) or a direct digital controller (DDC).

[0177] It also includes: a pressure sensor, which is electrically connected to the controller and is used to detect the air pressure value in the drying room;

[0178] It also includes: cloud servers, which are used for online training of MPC models;

[0179] Cloud server types include, but are not limited to, single servers, server clusters, cloud-based servers, and cloud-based server clusters; server configurations should meet the following requirements:

[0180] 8-core CPU, 32GB RAM, 1TB storage space;

[0181] High-speed internet connection bandwidth of 1Gbps;

[0182] Security measures include firewall configuration, encrypted data transmission, and regular backups.

[0183] It also supports basic cloud computing services such as network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0184] The cloud server is responsible for data processing, storage and analysis, as well as the deployment and optimization of the MPC model. It can access the MPC control system in real time through the cloud server. It performs deep learning and intelligent analysis on multi-source sensor data to make predictive maintenance, dynamic scheduling and adaptive control, thereby accurately adjusting parameters such as vehicle speed, fan speed, heat source power and exhaust opening, realizing intelligent control of fabric moisture content, and significantly improving control accuracy and production efficiency.

[0185] Experimental results:

[0186] Table 1: Product Weight Increase

[0187]

[0188] Table 2: Enhanced Enterprise Competitiveness

[0189] index Traditional methods Method of the present invention Economic benefits Equipment maintenance costs Annual maintenance cost is approximately 80,000 yuan (per unit). Annual maintenance cost is approximately 60,000 yuan. Reduced by 25% Response speed Control cycle 100ms Control cycle ≤ 50ms Increase by 50% Production transition time (cotton → synthetic fiber) 8-10 minutes 3 minutes Increased by 62% High-end product development cycle 40-60 days 24–36 days shorten by 40% Customer order fulfillment rate 85％ 95％ Increase by 12%

[0190] Table 3: Contributions of Green Production

[0191]

[0192]

[0193] Table 4: Comparison of Key Technical Parameters

[0194]

[0195] A dyeing and finishing company has achieved remarkable results in rapid production using the method of this invention;

[0196] Taking cotton fabrics as an example, data shows that after implementing the system, the difference in transverse moisture content of the fabric decreased from ±3% to within ±1%; each batch of production previously took about 8 hours, but now it only takes 6 hours. This improvement in production efficiency is attributed to the system's precise control over fabric moisture content and intelligent optimization of the production process. Furthermore, due to improved product quality and shorter delivery times, the company has gained more customer trust and orders.

[0197] After implementing this system, precise control over fabric moisture content was achieved, reducing downtime and adjustment time caused by substandard moisture content, and improving the continuity and stability of the production line. Data shows that the overall equipment efficiency (OEE) increased from 75% to over 85%, bringing significant economic benefits to the company.

[0198] This invention significantly improves product quality. First, it ensures the stability of physical properties and precise moisture content control, guaranteeing that the fabric's moisture regain meets process requirements (e.g., 8%–10% for cotton and 4%–6% for synthetic fibers), thus avoiding fiber embrittlement due to over-drying or mildew caused by under-drying.

[0199] The combined effect of humidity gradient constraint and air pressure control improves fabric width stability by 30% and controls weight deviation within ±1.5g / m². 2 within;

[0200] Multivariate model prediction and dynamic optimization form a differentiated advantage, which is suitable for the production of high-end fabrics (such as functional home textiles and sportswear) and meets customers' stringent requirements for moisture content of ±0.5%.

[0201] Appearance quality optimization: Dynamically adjust vehicle speed and heat source power to reduce color variations and color fastness degradation caused by temperature fluctuations; for example, in polyester dyed fabrics, the color difference value is reduced from 1.2 to below 0.8; the air duct blockage warning mechanism (ΔTi,i+1>10℃ for 10 seconds) can avoid local high temperature damage and reduce the defect rate by 0.5% to 1%.

[0202] This invention effectively improves production efficiency and controls production costs; optimizes equipment operation, limits the output of actuators (e.g., vehicle speed ΔM±2m / min, fan ΔF±3Hz) and provides interlock protection (heat source drops to 20% at low vehicle speeds), extending equipment life by more than 20% and reducing maintenance costs; the SQP algorithm solves in ≤50ms, improving response speed by 50% and supporting rapid production changeovers (e.g., switching from cotton to synthetic fibers takes only 3 minutes);

[0203] Energy efficiency is improved by coordinating the regulation of heat sources and fans based on model predictions, resulting in energy savings of 15% to 25% compared to traditional PID control; for example, when processing 5,000 meters / day of fabric, it can save approximately 120,000 kilowatt-hours of electricity per year.

[0204] This invention is highly intelligent and adaptable, adaptively adjusting process parameters and dynamically adjusting prediction model parameters (e.g., α = 0.025–0.035, β = 0.10–0.15) to automatically adapt to different weights (50–500 g / m³). 2 Fabrics with a width of 1.5–3.2 m can reduce human intervention.

[0205] Cloud servers support deep learning from historical data, optimize process formula libraries, and shorten the new product development cycle by 40%.

[0206] Fault warning and diagnosis, abnormal temperature gradient detection and exhaust valve linkage adjustment can respond to air duct blockage within 30 seconds, avoiding downtime losses due to equipment failure (such as a downtime loss of about 2,000 yuan per hour for a single unit).

[0207] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for controlling the moisture content of fabric dropped from a stenter based on MPC, characterized in that, Includes the following steps: Step 1: Collect control parameters; Step 2: Construct the MPC model. The MPC model includes: taking the measured and preset moisture content and the first control parameter state vector as inputs, and outputting the predicted moisture content; aiming to minimize the difference between the predicted moisture content and the preset moisture content; solving and optimizing the control increment using the optimization matrix and sequential quadratic programming; and using the optimal control increment to regulate the action of the actuator. Step two specifically includes: Preset prediction time domain N p Control time domain N c W H and W u ; Construct the prediction equation: ; in, (k+1) represents the predicted moisture content at time k+1; (k) represents the measured moisture content of the fabric at the point where it is laid at time k; α is the temperature influence coefficient; β is the humidity self-regulation coefficient; γ is the hot air ratio coefficient; ΔT is the difference in average temperature between adjacent times. is the adjustable coefficient; v is the vehicle speed; Q / F(k) is the ratio of heat source power to circulating fan flow rate; Construct the minimum objective function: ; Among them, W H To weigh the error weights; W u Weighting the control increment smoothness; To control the increment; Construct an optimization matrix and use the SQP algorithm to solve for Δu; The optimization of the control increment employs online correction of model parameters, with the formula: θ(k'+1)=θ(k')+μ·J －1 (θ(k')) (H(k')-H pred (k')); where μ is the adaptive learning rate, θ(k') is the control parameter coefficient vector at the k'th iteration, and J - ¹ represents the inverse of the second Jacobian matrix, H(k') is the measured moisture content of the fabric at the point where it is laid during the k'th iteration, and H... pred (k') Predicted water content at the k'+1th iteration.

2. The method for controlling the moisture content of fabric dropped from a stenter based on MPC according to claim 1, characterized in that, The formula for the optimization matrix is: ＝ ； Where A is the first Jacobian matrix, b H b u Let ΔX be the vector of error term and control increment term, and let ΔX represent the increment of state vector X(k).

3. The method for controlling the moisture content of fabric dropped from a stenter based on MPC according to claim 2, characterized in that, The control increment is constrained by the humidity gradient between adjacent drying rooms.

4. The method for controlling the moisture content of fabric dropped from a stenter based on MPC according to claim 2, characterized in that, The control increment is constrained by the air pressure difference between adjacent drying rooms.

5. The method for controlling the moisture content of fabric dropped from a stenter based on MPC according to claim 1, characterized in that, The control increments include: the change in the speed of the vehicle speed motor, the change in the flow rate of the circulating fan at adjacent time intervals, the change in the power of the heat source at adjacent time intervals, and the change in the exhaust volume at adjacent time intervals.

6. The method for controlling the moisture content of fabric dropped from a stenter based on MPC according to claim 1, characterized in that, The control parameters include: temperature, humidity, moisture content, air pressure difference, vehicle speed, exhaust volume, and the ratio of heat source power to circulating fan flow rate.

7. The method for controlling the moisture content of fabric dropped from a stenter based on MPC according to claim 1, characterized in that, The first control parameters include: average humidity of the drying room, average temperature of the drying room, vehicle speed, and the ratio of heat source power to circulating fan flow rate.

8. A system employing the MPC-based method for controlling the moisture content of fabric dropped from a stenter as described in any one of claims 1-7, characterized in that, include: Actuators, controllers, humidity sensors, temperature sensors, and moisture content sensors; in, Humidity sensors are used to collect humidity values ​​in each drying room; Temperature sensors are used to collect temperature values ​​inside each drying chamber; Moisture content sensors are used to detect the moisture content of fabric at the point where it is laid. The controller uses the collected humidity, temperature, and moisture content values, along with a trained MPC model, to control the actions of the actuators and achieve moisture content regulation.

Images